X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Fstructure.c;h=cb79f9996606b8209f0913267d0a3aed250a7370;hp=466c46624dda89ce1dafcf2fef8178ba3ddc2bab;hb=f3d9ef5ccb0f028fb2152d3df34bf5fdada4c350;hpb=0cc60c89bb2ee51d5313ec54e63157b11f399633 diff --git a/pctb/structure.c b/pctb/structure.c index 466c466..cb79f99 100644 --- a/pctb/structure.c +++ b/pctb/structure.c @@ -27,7 +27,7 @@ #include "structure.h" -static const CanonImage *cim; +static CanonImage *cim; static inline char get(int x, int y) { return cim->d[y * cim->w + x]; } static inline char get_p(Point p) { return get(p.x,p.y); } @@ -58,7 +58,8 @@ const CanonColourInfo canoncolourinfos[]= { { 0x6B828C, '*' }, /* background of ship status meter area */ { 0x934405, '*' }, /* border of ship meter area */ { 0x7D9094, '+' }, /* interbox */ - { 0x022158, 'a' }, /* ahoy /w output foreground */ + { 0x022158, 'O' }, /* ahoy /w output foreground */ + { 0xB5B686, 'H' }, /* ahoy /w output heading background */ { 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */ { 0xADB5AF, ' ' }, /* background - dark */ @@ -91,10 +92,15 @@ const CanonColourInfo canoncolourinfos[]= { static void mustfail1(const char *file, int line, const char *what) { fprintf(stderr, - "\n\n" - "Unable to figure out contents of YPP client display.\n" - " Check that your client is logged in has the correct display.\n" - " If that isn't the problem, please report this as a fault.\n\n" + "\n\n" + "Unable to figure out contents of YPP client display.\n" + "Please check the following:\n" + " * YPP client is showing commodity listing screen\n" + " * YPP client window is on top (we try to raise it but your window\n" + " manager might have prevented that from succeeding)\n" + " * Your X display is 24bpp (NB some VNC servers use 16bpp by default)\n" + "\n" + "If all of these are true, please report this as a fault.\n\n" "Technical details:" " %s:%d: requirement failed:\n" " %s\n", @@ -116,8 +122,11 @@ static void mustfail2(void) { #define MP(v) fprintf(stderr," %s=%d,%d",#v,(v).x,(v).y) #define MI(v) fprintf(stderr," %s=%d", #v,(v)) +#define MIL(v) fprintf(stderr," %s=%ld", #v,(v)) +#define MRGB(v) fprintf(stderr," %s=%06lx", #v,(v)) #define MC(v) fprintf(stderr," %s='%c'", #v,(v)) #define MS(v) fprintf(stderr," %s=\"%s\"", #v,(v)) +#define MF(v) fprintf(stderr," %s=\"%f\"", #v,(v)) #define MSB(v) fprintf(stderr," %s", (v)) #define MR(v) fprintf(stderr," %s=%d,%d..%d,%d",\ #v,(v).tl.x,(v).tl.y,(v).br.x,(v).br.y) @@ -126,20 +135,23 @@ static void mustfail2(void) { #define REQUIRE_RECTANGLE(tlx,tly,brx,bry,ok) \ require_rectangle(tlx, tly, brx, bry, ok, __LINE__); -static void require_rectangle(int tlx, int tly, int brx, int bry, - const char *ok, int lineno) { +#define FOR_P_RECT(p,rr) \ + for ((p).x=(rr).tl.x; (p).x<=(rr).br.x; (p).x++) \ + for ((p).y=(rr).tl.y; (p).y<=(rr).br.y; (p).y++) + +static void require_rectangle_r(Rect rr, const char *ok, int lineno) { Point p; - for (p.x=tlx; p.x<=brx; p.x++) - for (p.y=tly; p.y<=bry; p.y++) { - int c= get_p(p); - MUST( strchr(ok,c), ({ - Rect rm={{tlx,tly},{brx,bry}}; - MI(lineno),MR(rm);MP(p);MS(ok); - })); - } + FOR_P_RECT(p,rr) { + int c= get_p(p); + MUST( strchr(ok,c), ({ + MI(lineno),MR(rr);MP(p);MS(ok); + })); + } } -static void require_rectangle_r(Rect rr, const char *ok, int lineno) { - require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, ok, lineno); +static void require_rectangle(int tlx, int tly, int brx, int bry, + const char *ok, int lineno) { + Rect rr= {{tlx,tly},{brx,bry}}; + require_rectangle_r(rr, ok, lineno); } static void debug_rect(const char *what, int whati, Rect rr) { @@ -198,23 +210,23 @@ static int commod_selector_matches(Rect search, const char *const *all, ); \ }while(0) -#define ADJUST_BOX(search,insidechrs,want, lim,LIMIT_MUST, TLBR,XY,increm) \ - for (;;) { \ - LIMIT_MUST( (search).tl.XY != (search).br.XY && \ - (search).tl.XY != (lim), \ - MR((search));MSB(#TLBR);MSB(#XY) ); \ - int got=0; \ - Point p=(search).tl; \ - for (p.XY=(search).TLBR.XY; \ - p.OTHERCOORD_##XY <= (search).br.OTHERCOORD_##XY; \ - p.OTHERCOORD_##XY++) \ - got += !!strchr(insidechrs, get_p(p)); \ - if (got >= (want)) \ - break; \ - (search).TLBR.XY += increm; \ +#define ADJUST_BOX(search,insidechrs,OP,want, lim,LIMIT_MUST, TLBR,XY,increm) \ + for (;;) { \ + LIMIT_MUST( (search).tl.XY != (search).br.XY && \ + (search).tl.XY != (lim), \ + MR((search));MSB(#TLBR);MSB(#XY) ); \ + int got=0; \ + Point p=(search).tl; \ + for (p.XY=(search).TLBR.XY; \ + p.OTHERCOORD_##XY <= (search).br.OTHERCOORD_##XY; \ + p.OTHERCOORD_##XY++) \ + got += !!strchr(insidechrs, get_p(p)); \ + if ((got) OP (want)) \ + break; \ + (search).TLBR.XY += increm; \ } -void find_structure(const CanonImage *im, int *max_relevant_y_r) { +void find_structure(CanonImage *im, int *max_relevant_y_r) { cim= im; Rect whole = { {0,0}, {cim->w-1,cim->h-1} }; @@ -307,13 +319,15 @@ void find_structure(const CanonImage *im, int *max_relevant_y_r) { SET_ONCE(text_h, comminty - 1); if (max_relevant_y_r) SET_ONCE(*max_relevant_y_r, mainr.br.y + 10); + + MUST( text_h <= OCR_MAX_H, MI(text_h) ); } void check_correct_commodities(void) { Rect search= { { 50,39 }, { 130,59 } }; - ADJUST_BOX(search,"_",10, cim->h, MUST, tl,y,+1); - ADJUST_BOX(search,"_",10, 0, MUST, br,y,-1); + ADJUST_BOX(search,"_",>=,10, cim->h, MUST, tl,y,+1); + ADJUST_BOX(search,"_",>=,10, 0, MUST, br,y,-1); debug_rect("commodselr",1, search); @@ -368,16 +382,12 @@ static void file_read_image_ppm(FILE *f) { struct pam inpam; unsigned char rgb_buf[3]; CanonImage *im; - RgbImage *ri=0; pnm_readpaminit(f, &inpam, sizeof(inpam)); if (!(inpam.maxval == 255 && inpam.bytes_per_sample == 1 && inpam.format == RPPM_FORMAT)) - fatal("PNM screenshot(s) file must be 8bpp 1 byte per sample RGB"); - - if (!npages) - page0_rgbimage= ri= alloc_rgb_image(inpam.width, inpam.height); + fatal("PNM screenshot(s) file must be 8bpp 1 byte-per-sample RGB raw"); CANONICALISE_IMAGE(im, inpam.width, inpam.height, { int r= fread(&rgb_buf,1,3,f); @@ -389,8 +399,6 @@ static void file_read_image_ppm(FILE *f) { ((unsigned long)rgb_buf[1]<<8) | (rgb_buf[2]); - if (ri) - CANONIMG_ALSO_STORERGB(ri); }); sysassert(!ferror(screenshot_file)); @@ -443,13 +451,85 @@ static void find_commodity(int offset, Rect *rr) { REQUIRE_RECTANGLE(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+"); } -static void find_table_entry(Rect commod, int colno, Rect *cellr) { - cellr->tl.y= commod.tl.y; - cellr->br.y= commod.br.y; - cellr->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2; - cellr->br.x= colrightx[colno]; - debug_rect("cell", colno, *cellr); - require_rectangle_r(*cellr, " o", __LINE__); +static void find_table_entry(Rect commod, int colno, Rect *cell) { + cell->tl.y= commod.tl.y; + cell->br.y= commod.br.y; + cell->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2; + cell->br.x= colrightx[colno]; + debug_rect("cell", colno, *cell); + + const RgbImage *ri= cim->rgb; + + Rgb background= ri_rgb(ri, cell->br.x, cell->br.y); + long bg_count=0, light_count=0, dark_count=0; + Point p; + FOR_P_RECT(p,*cell) { + Rgb here= ri_rgb(ri, p.x, p.y); + if (here == background) bg_count++; + else if (here < background) dark_count++; + else if (here > background) light_count++; + } + long total_count= RECT_W(*cell) * RECT_H(*cell); + MUST( bg_count > total_count / 2, + MR(*cell);MIL(total_count);MIL(bg_count); + MIL(light_count);MIL(dark_count) ); + if (bg_count == total_count) + return; + + MUST( !!dark_count != !!light_count, + MR(*cell);MIL(total_count);MIL(bg_count); + MIL(light_count);MIL(dark_count) ); + + debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld light=%ld dark=%ld\n", + colno, cell->tl.x,cell->tl.y, cell->br.x,cell->br.y, + bg_count, light_count, dark_count); + + Rgb foreground; + double fg_extra; + if (light_count) { + foreground= 0xffffffU; + fg_extra= +1; + } else { + foreground= 0; + fg_extra= -1; + } + int monochrome= 1; + + FOR_P_RECT(p,*cell) { + Rgb here= ri_rgb(ri, p.x, p.y); + double alpha[3], alpha_mean=0; + int i; + for (i=0; i<3; i++) { + unsigned char here_chan= here >> (i*8); + unsigned char bg_chan= background >> (i*8); + unsigned char fg_chan= foreground >> (i*8); + double alpha_chan= + ((double)here_chan - (double)bg_chan) / + ((fg_chan + fg_extra) - (double)bg_chan); + alpha[i]= alpha_chan; + alpha_mean += alpha_chan * (1/3.0); + } + + double thresh= 1.5/AAMAXVAL; + double alpha_min= alpha_mean - thresh; + double alpha_max= alpha_mean + thresh; + for (i=0; i<3; i++) + MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max, + MI(i);MRGB(here);MRGB(background);MRGB(foreground); + MF(alpha_min);MF(alpha[i]);MF(alpha_max) ); + + MUST( 0 <= alpha_mean && alpha_mean < 0.999, + MRGB(here);MRGB(background);MRGB(foreground); + MF(alpha_mean); MF(alpha[0]);MF(alpha[1]);MF(alpha[2]); ); + int here_int= floor((AAMAXVAL+1)*alpha_mean); + assert(here_int <= AAMAXVAL); + if (!(here_int==0 || here_int==AAMAXVAL)) monochrome=0; + cim->d[p.y * cim->w + p.x]= '0' + here_int; + } + + debug_rect("cell0M", colno, *cell); + + require_rectangle_r(*cell, "0123456789", __LINE__); } static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) { @@ -459,21 +539,17 @@ static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) { Pixcol cols[w+1]; int x,y; for (x=0; x= '0' && pixel <= '0'+AAMAXVAL, + MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); ); + pixcol_p_add(&cols[x], y, pixel-'0'); } - cols[x]= cx; } - cols[w]= 0; + FILLZERO(cols[w]); results= ocr(rd,ct,w,cols); for (res=results; res->s; res++) @@ -522,38 +598,44 @@ void analyse(FILE *tsv_output) { //static Rect islandnamer; -void find_islandname(void) { - Rect sunshiner; - char sunshine[MAXIMGIDENT], archisland[MAXIMGIDENT]; +DEBUG_DEFINE_SOME_DEBUGF(structcolon,colondebugf) - RgbImage *ri= alloc_rgb_image(page0_rgbimage->w, page0_rgbimage->h); - const unsigned char *srcp; - unsigned char *destp, *endp; - for (srcp=page0_rgbimage->data, destp=ri->data, - endp= ri->data + 3 * ri->w * ri->h; - destp < endp; - srcp++, destp++) { - unsigned char c= *srcp & 0xf0; - *destp= c | (c>>4); - } +Rect find_sunshine_widget(void) { + Rect sunshiner; - cim= page_images[0]; - sunshiner.tl.x= cim->w - 1034 + 885; sunshiner.br.x= cim->w - 1034 + 1020; sunshiner.tl.y= 227; sunshiner.br.y= 228; - ADJUST_BOX(sunshiner,"o*",30, 100,MUST, tl,y,-1); - ADJUST_BOX(sunshiner,"o*",30, 100,MUST, br,y,+1); + ADJUST_BOX(sunshiner,"o*",>=,30, 100,MUST, tl,y,-1); + ADJUST_BOX(sunshiner,"o*",>=,30, 100,MUST, br,y,+1); debug_rect("sunshiner",0, sunshiner); MUST(sunshiner.br.y - sunshiner.tl.y > 20, MR(sunshiner)); sunshiner.br.y--; - ADJUST_BOX(sunshiner,"o",20, (cim->w - 1034 + 700), MUST, tl,x,-1); - ADJUST_BOX(sunshiner,"o",20, cim->w, MUST, br,x,+1); + ADJUST_BOX(sunshiner,"o",>=,20, (cim->w - 1034 + 700), MUST, tl,x,-1); + ADJUST_BOX(sunshiner,"o",>=,20, cim->w, MUST, br,x,+1); debug_rect("sunshiner",1, sunshiner); + return sunshiner; +} + +void find_islandname(RgbImage *ri) { + find_structure(page_images[0], 0); + + Rect sunshiner= find_sunshine_widget(); + char sunshine[MAXIMGIDENT], archisland[MAXIMGIDENT]; + + const unsigned char *srcp; + unsigned char *destp, *endp; + for (srcp=page_images[0]->rgb->data, destp=ri->data, + endp= ri->data + 3 * ri->w * ri->h; + destp < endp; + srcp++, destp++) { + unsigned char c= *srcp & 0xf0; + *destp= c | (c>>4); + } identify_rgbimage(ri, sunshiner, sunshine, "sunshine widget"); @@ -565,11 +647,11 @@ void find_islandname(void) { islandnamer.tl.y= 128; islandnamer.br.y= 156; - ADJUST_BOX(islandnamer,"o",5, 0, MUST, tl,y,+1); - ADJUST_BOX(islandnamer,"o",5, cim->h, MUST, br,y,-1); + ADJUST_BOX(islandnamer,"o",>=,5, 0, MUST, tl,y,+1); + ADJUST_BOX(islandnamer,"o",>=,5, cim->h, MUST, br,y,-1); - ADJUST_BOX(islandnamer,"o",1, 0, MUST, tl,x,+1); - ADJUST_BOX(islandnamer,"o",1, cim->w, MUST, br,x,-1); + ADJUST_BOX(islandnamer,"o",>=,1, 0, MUST, tl,x,+1); + ADJUST_BOX(islandnamer,"o",>=,1, cim->w, MUST, br,x,-1); debug_rect("islandnamer",0, islandnamer); // int larger_islandnamebry= islandnamer.tl.y + 25; @@ -588,15 +670,107 @@ void find_islandname(void) { } } + identify_rgbimage(ri, islandnamer, archisland, "island"); + } else if (!strcmp(sunshine,"Land - Ahoy!")) { + Rect islandnamer; + + islandnamer.tl.x= (sunshiner.tl.x + sunshiner.br.x) / 2; + islandnamer.tl.y= sunshiner.tl.y + 100; + islandnamer.br= islandnamer.tl; + debug_rect("islandnamer",__LINE__, islandnamer); + + WALK_UNTIL_MUST(islandnamer.tl,y, -1, sunshiner.br.y, 'H'); + WALK_UNTIL_MUST(islandnamer.tl,x, -1, 0, 'o'); + WALK_UNTIL_MUST(islandnamer.br,x, +1, cim->w, 'o'); + debug_rect("islandnamer",__LINE__, islandnamer); + +#define RW (RECT_W(islandnamer)) +#define RH (RECT_H(islandnamer)) + + ADJUST_BOX(islandnamer,"O",>=,RW-4, cim->h, MUST,br,y,+1); + debug_rect("islandnamer",__LINE__, islandnamer); + + islandnamer.br.y += 2; + + ADJUST_BOX(islandnamer,"*",<,RW, cim->h, MUST,br,y,+1); + debug_rect("islandnamer",__LINE__, islandnamer); + + islandnamer.tl.y= islandnamer.br.y-1; + islandnamer.br.y= islandnamer.br.y+1; + debug_rect("islandnamer",__LINE__, islandnamer); + + ADJUST_BOX(islandnamer,"*",>=,RW, cim->h, MUST,br,y,+1); + debug_rect("islandnamer",__LINE__, islandnamer); + + ADJUST_BOX(islandnamer,"*",<, RH, cim->w, MUST,tl,x,+1); + debug_rect("islandnamer",__LINE__, islandnamer); + + MUST( RECT_H(islandnamer) <= 30, MR(islandnamer)); + + Point p; + int nspaces=1, might_be_colon=0; + uint32_t colon_pattern= 0; + p.y=-1; + + for (p.x=islandnamer.br.x; p.x>islandnamer.tl.x; p.x--) { + colondebugf("structcolon: x=%4d nsp=%2d mbc=%d cp=%08"PRIx32" ", + p.x, nspaces, might_be_colon, colon_pattern); + + uint32_t pattern=0; + int runs[32], nruns=0; + runs[0]=0; runs[1]=0; + + for (p.y=islandnamer.tl.y; p.y<=islandnamer.br.y; p.y++) { + pattern <<= 1; + switch (get_p(p)) { + case '*': if (runs[nruns]) { nruns++; runs[nruns]=0; } break; + case 'O': runs[nruns]++; pattern |= 1u; break; + default: pattern |= 1UL<<31; break; + } + } + + colondebugf(" pat=%08"PRIx32" nruns=%d runs[]={%d,%d..} ", + pattern, nruns, runs[0],runs[1]); + + if (!pattern) { + if (might_be_colon) + /* omg it _is_ a colon */ + goto colon_found; + nspaces++; + might_be_colon=0; + } else { + if (!(pattern & 1UL<<31) && nruns==2 && runs[1]==runs[0]) { + if (!nspaces) { + if (pattern==colon_pattern) + goto ok_might_be_colon; + } else if (nspaces>=2) { + colon_pattern= pattern; + might_be_colon=1; + goto ok_might_be_colon; + } + } + might_be_colon=0; + ok_might_be_colon: + nspaces= 0; + } + colondebugf(" nsp=%2d mbc=%d\n", nspaces, might_be_colon); + } + MUST(!"colon found", MP(p);MR(islandnamer) ); + + colon_found: + colondebugf(" found\n"); + islandnamer.br.x= p.x; + identify_rgbimage(ri, islandnamer, archisland, "island"); } else { - assert(!"not vessel"); + + MUST(!"sunshine shows ship or ahoy", MS(sunshine) ); + } char *delim= strstr(archisland," - "); assert(delim); - archipelago= masprintf("%.*s", delim-archisland, archisland); + archipelago= masprintf("%.*s", (int)(delim-archisland), archisland); island= masprintf("%s", delim+3); - free(ri); }